Lightning-AI / pytorch-lightning UNCLAIMED

Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.

0 0 0 Python

Fix: Allow trainer to accept CUDAAccelerator instance as accelerator with FSDP strategy (#20964)

* Add test for FSDP with CUDAAccelerator instance
* update error message
* update the test

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
(cherry picked from commit 460c60cc970aa95f9ba17d0b9007da3ce28e2489)

Bhimraj Yadav committed 8mo ago

b8bd96f56dee0d3e6a0410dcb0c6d17e325eff5e

Parent: 9c6a8e0

Committed by Jirka Borovec <[email protected]> on 8/13/2025, 7:19:50 PM